AI Cannot Repair Weak Commercial Data

Evidence context

The CIPS/GEP AI-readiness study reports that all 180 senior supply-chain leaders surveyed had access to AI, but only 5% had scaled it into operations. Deloitte finds that Digital Masters combine technology investment with talent and operating-model change, achieving stronger procurement outcomes than followers.

The gap between access and scale is highly relevant to FM. Contract documents, asset records, supplier names, scopes, work orders and invoices are often inconsistent across sites. An AI tool may classify or summarise this information quickly, but it cannot make ambiguous obligations or missing records authoritative.

Osmos Global analysis

High-value use cases start with bounded decisions: identifying duplicate invoices, comparing stated exclusions, flagging expiring certifications or summarising supplier performance. Each requires a controlled source set, human review and a defined consequence for error.

AI governance should distinguish assistance from authority. Drafting a comparison is different from approving a supplier, changing a maintenance plan or accepting a compliance record. The control environment should reflect that difference.

Scaling should follow evidence of accuracy, adoption and business value. A pilot that produces an impressive demonstration without integration, ownership or monitoring is not operational transformation.

Leadership implications

For FM and CRE leaders, the issue is governance before mechanics: establish the operating outcome, evidence standard and risk boundary before choosing the commercial mechanism. Procurement leaders should ensure the evaluation model makes lifecycle value and uncertainty visible. Providers should be asked to demonstrate how their proposed method changes decisions, not merely how it produces reports.

Implementation guidance

Start with one service or decision where current performance can be reconstructed. Record the baseline, ownership, data source, approval route and foreseeable failure modes. Pilot the proposed commercial control, review exceptions with frontline teams, and scale only after the evidence survives finance, technical and user scrutiny. This creates a repeatable operating discipline instead of a one-off sourcing event.

Practical actions

• Prioritise narrow, auditable use cases. • Clean contract and supplier master data first. • Keep humans accountable for consequential decisions. • Measure exception rates and realised value after deployment.

Risks, limitations and unresolved questions The CIPS/GEP study is cross-industry and the public page provides limited methodology detail. AI results should not be generalised without use-case testing.

Source notes

[1] CIPS / GEP. The Supply Chain AI Readiness Report. 1 June 2026. https://cips-download.cips.org/expert-reports/gep-ai-readiness-report [2] JLL. Global State of Facilities Management Report 2025. 12 November 2025. https://www.jll.com/en-us/insights/global-state-of-facilities-management-report